Calculating Workplace WBGT from Meteorological Data: A Tool for

Transcription

Calculating Workplace WBGT from Meteorological Data: A Tool for
Review Article
Industrial Health 2012, 50, 267–278
Calculating Workplace WBGT from Meteorological
Data: A Tool for Climate Change Assessment
Bruno LEMKE1* and Tord KJELLSTROM2, 3
1
School of Health, Nelson Marlborough Institute of Technology, New Zealand
National Centre for Epidemiology and Population Health, Australian National University, Australia
3
Centre for Global Health Research, Umea University, Sweden
2
Received November 25, 2011 and accepted May 7, 2012
Published online in J-STAGE May 30, 2012
Abstract: The WBGT heat stress index has been well tested under a variety of climatic conditions
and quantitative links have been established between WBGT and the work-rest cycles needed to
prevent heat stress effects at the workplace. While there are more specific methods based on individual physiological measurements to determine heat strain in an individual worker, the WBGT
index is used in international and national standards to specify workplace heat stress risks. In order
to assess time trends of occupational heat exposure at population level, weather station records or
climate modelling are the most widely available data sources. The prescribed method to measure
WBGT requires special equipment which is not used at weather stations. We compared published
methods to calculate outdoor and indoor WBGT from standard climate data, such as air temperature, dew point temperature, wind speed and solar radiation. Specific criteria for recommending
a method were developed and original measurements were used to evaluate the different methods.
We recommend the method of Liljegren et al. (2008) for calculating outdoor WBGT and the method
by Bernard et al. (1999) for indoor WBGT when estimating climate change impacts on occupational
heat stress at a population level.
Key words: Heat stress, WBGT, Workplace, Weather data, Indoor, Outdoor, Climate change
Introduction
Measuring the effects of heat exposure in occupational
health has many perspectives: for instance, the effect of
heat on individuals undertaking a particular activity; or the
effect of heat on a population of workers in a particular
region; or the effect of increasing heat exposure due to
climate change on the working population in different
regions. Measurement of heat effects fall into two categories: heat stress and heat strain. Conventional engineering
*To whom correspondence should be addressed.
E-mail: [email protected]
©2012 National Institute of Occupational Safety and Health
terminology defines stress as external forces and strain as
the response by the object or individual to those stresses.
In this paper we focus on heat stress for which numerous indexes have been defined 1) . We will use WBGT
(Wet Bulb Globe Temperature) because that is a wellestablished heat index for workplace applications with
recommended rest/work cycles at different metabolic rates
clearly specified in an international standard2).
New heat stress indexes such as the UTCI (Universal
Thermal Climate Index) are based on the heat balance
mechanisms of the human body3), and although they are
based on the best physiological models of the body’s
response to heat, they are not as practical as the WBGT index. For example they do not take into account differences
in metabolic rates during work, or the impact of special-
268
B LEMKE et al.
Table 1. Often used abbreviations (all units in °C unless otherwise shown)
WBGTid
Indoor Wet bulb globe temperature
Td
Dew point temperature
Atmospheric pressure (hPa)
WBGTod
Outdoor Wet bulb globe temperature
Pa
Ta
Air temperature
ρ
Water vapour pressure (hPa)
Tnwb
Natural wet bulb temperature
SR
Solar Radiation (w/m 2)
Tpwb
Psychrometric wet bulb temperature
v
Wind speed (m/s)
Tg
Black globe temperature
ized protective clothing, or the constant change in position
and movement during real work situations.
If individual heat strain data is required then physiological models exist to calculate heat strain. These include
the required sweat rate model4), the Predicted Heat Strain
(PHS) Model5), the USAF model6) and the Fiala model7)
to name a few.
The core body temperature of all humans is maintained
close to 37°C. The main mechanism of internal heat gain is
the heat generated by muscles that work at approximately
20% efficiency8). Heat can be transferred to/from the body
by convection, conduction, radiation and evaporation of
water (sweat). Environmental factors that influence these
heat transfer mechanisms, and the resulting heat stress, are
air temperature, wind speed, humidity and heat radiation
sources8). The heat stress is also dependent on the clothing
and the intensity of muscular work (the metabolic rate) of
the person8).
Geographic variations of personal factors can be specified by a heat exposure standard that can vary from country to country. Such standards translate the WBGT index
to a health risk function for the population 9, 10). Indeed,
the heat stress standard is different in cooler countries (e.g.
England) than in hot countries (e.g. India) where the population is more acclimatized and better adapted to higher
temperatures.
Increased heat exposure raises the core body temperature of the human body. While some increase in core
temperature above 37°C is acceptable, an increase beyond
39°C creates health risks8), which vary from person to person, depending on ethnic group, age, gender, the duration
of high heat exposure, and the degree of acclimatization11).
The International Standard for heat stress uses WBGT
to recommend work-rest limits for work in hot environments2) in order to ensure that average core body temperatures of worker populations does not exceed 38°C. Many
countries have national standards based on this international standard for WBGT limit values9).
The WBGT index was developed after detailed studies by US military ergonomists in the 1950s12, 13). While
WBGT has its critics14–16), it has all the correct components to indicate physiological heat stress8). One criticism
of WBGT17, 18) is that it is too stringent for determining
conditions when full work load should be reduced. This
criticism has less to do with the WBGT index than the
ISO standard criteria for reduction of the work load2). For
instance, Nag et al.17) considered that an increase in core
body temperature up to 39°C was acceptable, while ISO
limits the increase to 38°C. The international standard is
currently being updated by ISO19) to address some of these
criticisms and to update the scientific source materials.
A major disadvantage of WBGT is that, until recently, it
could not be easily calculated from standard meteorological data. In this paper we review, assess and compare different published methods to calculate WBGT from readily
available meteorological data, namely temperature, humidity, wind speed and solar radiation. Our main objective is
to identify and develop a valid global method to calculate
current and future heat stress using weather station data,
and provide improved estimates of climate change related
variations in occupational heat stress.
Components in Calculations of WBGT
The heat exposure index WBGT (unit=°C) is a combination of the natural wet bulb temperature (Tnwb, measured with a wetted thermometer exposed to the wind and
heat radiation at the site), the black globe temperature (Tg,
measured inside a 150 mm diameter black globe), and the
air temperature (Ta, measured with a “normal” thermometer shaded from direct heat radiation). Table 1 presents a
summary of variables discussed in this article.
Equation 18): outdoors (in conditions of direct short wave
radiation): WBGTod = 0.7 Tnwb + 0.2 Tg + 0.1 Ta
Equation 28): indoors or outdoors in the shade (no direct
short wave radiation): WBGTid = 0.7 Tnwb + 0.3 Tg
These simple equations were developed more than 50 yr
ago12) and are still in use2). The wet bulb thermometer
simulates the cooling of the body via sweat evaporaIndustrial Health 2012, 50, 267–278
WBGT CALCULATIONS FROM METEOROLOGICAL DATA
tion, which is strongly related to air humidity. The globe
thermometer simulates the heat absorption from radiation
(from the sun or heat sources in the workplace). The
temperatures recorded by these two thermometers are also
modified by the air temperature (Ta) and the air movement
(wind speed) around them.
Different wet bulb temperatures
The natural wet bulb temperature (Tnwb) is the largest
component (70%) of WBGT (see Equations 1 and 2).
There are three versions of the “wet bulb temperature”.
1.The Natural Wet Bulb temperature (Tnwb) is the temperature of a wetted thermometer bulb in the natural
environment of wind and sun. Tnwb is a combination
of air temperature and humidity but it is also influenced
by heat radiation and wind speed.
2.The Wet Bulb temperature (Twb) used at some weather
stations is the wetted bulb left in natural wind conditions but shielded from direct sunlight (ie in the shade).
3.The psychrometric or thermodynamic wet bulb temperature (Tpwb) is used to calculate dew point and relative humidity. Here the wetted bulb is in the shade and
aspirated with a wind of 3−5 m/s created by a fan or by
rotating a wetted thermometer20, 21).
Modern meteorological data is usually based on the psychrometric wet bulb temperature (Tpwb) which is linked
to the dew point (Td) by the formula:
Equation 322): Td = 243.5 ln(ρ/6.112)/(17.67 − ln(ρ/6.112))
where ρ is the water vapor pressure in air (hecto-Pascals)
calculated with equation 4:
Equation 422): ρ = 6.112exp(17.67Tpwb/(Tpwb+243.5)) −
0.00066Patm(Ta-Tpwb)(1+ 0.00115Tpwb)
where Patm is the atmospheric pressure in hecto-Pascals
(hPa).
Low cost WBGT meters that use electronic components
to measure relative humidity rather than a wetted wick display the psychrometric wet bulb temperature rather than
the natural wet bulb temperature. For wind speeds above 3
m/s, it matters little which wet bulb temperature is used23).
Measurements made at low wind speed, resulted in the
natural wet bulb temperature being up to 1.5°C higher
than the psychrometric wet bulb temperature indoors and
nearly 10°C higher outdoors in the sun.
Globe temperature Tg
The globe temperature (Tg) is a combination of short
wave heat radiation (outdoors usually from the sun),
269
long wave radiation (outdoors usually from the soil),
and convective cooling due to wind on the thermometer.
It contributes 20–30% of the WBGT (Equations 1 and
2). While Tnwb is the largest component in the WBGT
formula (70%), the Tg can be up to three times higher than
Tnwb, so on balance both components can have a similar
influence on WBGT.
Low wind speeds have a considerable effect on Tg.
The heat gained by the globe from radiation is essentially
only lost by wind convection to cooler air around the
globe. If there is no wind then the temperature rise by a
stationary black globe left in the sun can be considerable.
However, humans rarely stand still when they are working
outdoors. Body movement generates air flow over the skin
so the “wind speed” on the skin will never be 0 (as for a
stationary WBGT monitor). In our calculations we use a
standard wind speed at 1 m/s, so for windless conditions
our calculated results will produce a more accurate WBGT
for a moving worker than measured by a stationary WBGT
meter. A standard wind speed of 1 m/s was chosen because
for actively working people, limb and torso movement
would create an apparent wind speed greater than this.
While WBGT shows a large dependence on wind speed
when the wind speed is low, once the wind speed is over
1 m/s there is only a minor increase in WBGT. For example, for an increase in wind speed from 1 m/s to 5 m/s
there is at most a 5% increase in WBGT.
The original WBGT meter with a 150 mm diameter
black globe takes approximately 20 min to reach equilibrium24), so modern WBGT meters often use a smaller
50 mm diameter black globe. The 50 mm globe of the
Quest Technology instrument reaches equilibrium in
10 min. Our calculations based on the “best” formula (see
later) showed that the smaller globe underestimates Tg by
less than 7% for 1 m/s wind speed and in direct sunlight.
Manufacturers of the smaller globes state that this has
been corrected in their WBGT readings25).
Criteria for Establishing a Valid WBGT
Calculation from Meteorological Data
Our criteria to determine the best methods and formulas
for calculating WBGT from meteorological data:
* The formula should only include well established meteorological variables related to temperature, humidity,
wind speed and solar radiation to calculate both Tnwb
and Tg as required by the WBGT formula (see equations 1 and 2).
* The method in deriving the formula should be based
270
on sound thermodynamic principles of heat exchange
between the environment and the black globe and the
wetted wick. Empirical values of constants may be included in the formula.
* The formula should cover all common conditions encountered in the outdoors (for WBGTod) or indoors (for
WBGTid).
* The formula should be easy to use.
* The formula should have been well tested with experimental data.
A number of authors have calculated WBGT from standard
meteorological data: Dernedde and Gilbert26) (1991), Bernard and Pourmoghani27) (1999), Hunter and Minyard28)
(1999), ABM 29) (Australian Bureau of Meteorology),
Tonouchi et al.30) (2006), Liljegren et al.31) (2008), Gaspar
and Quintela32) (2009). The methods are described below
and will be referred to using the first author’s name.
Published Methods for Calculating WBGT
and its Components
Dernedde and Bernard
Dernedde and Gilbert26) developed Tnwb temperature
formulas based on the heat exchange of a wetted wick in
the sun and wind. Bernard and Pourmoghani27) developed
this further resulting in the following equation where the
first term is the convective heat exchange, the second term
is the heat gained from radiation sources and the third term
is the heat lost by evaporation:
Equation 526, 27): h(Ta-Tnwb) + εw(SR-σ(Tnwb)4) − kQ(PsPw)/(Pa-Pw) = 0
Variables and constants: h is the heat transfer of convection, ε w is the emissivity of the wick, σ is the StefanBoltzmann constant, k is the mass transfer coefficient of
the wick, SR is solar radiation in W/m2, Q is the latent
heat of vaporization and Ps is the saturated partial pressure
at the wick temperature. Pw is the partial pressure of water
in the air and Patm is the atmospheric pressure. Ta and
Tnwb were defined in Equation 1.
Bernard et al.27) and Bernard23) use the principles of
heat exchange of a wetted wick with the environment and
actual measurements to derive a semi-empirical formula
for Tnwb for common summertime environmental conditions in the USA:
Equation 627):
a) When Tg is 4 °C higher than Ta; Tnwb = Tpwb + 0.25(Tg
− Ta) + 0.1/v1.1 − 0.2
B LEMKE et al.
b) When Tg − Ta is < 4 °C, and wind speed v is > 3 m/s,
Tnwb = Tpwb
c) Otherwise Tnwb = Ta − (0.96 + 0.069log10v) (Ta − Tpwb)
Unfortunately Bernard’s method does not include estimation of the black globe temperature in the sun. Their
theory and measurements all apply to indoor environments
without a solar radiation (SR) component. Hence this
method is not suitable for outdoor WBGT, but it would be
appropriate for indoor WBGT calculations.
For indoor conditions with no strong radiation sources,
the approximation that Tg = Ta was tested by us in the
course of indoor WBGT measurements and agreed to
within 0.5°C. So using this approximation with the indoor
WBGT formula (Equation 2) along with Bernard’s Tnwb
formula (Equation 6) we obtain:
Equation 7:WBGTid = 0.7Tpwb + 0.3Ta
(v > 3m/s; Tnwb = Tpwb; Tg = Ta)
WBGTid = 0.67Tpwb + 0.33Ta − 0.048 log10v (Ta − Tpwb)
(v = 0.3–3 m/s)
Wind speeds less than 0.3 m/s are not included in this
analysis because a working person is unlikely to be completely stationary so an apparent wind speed of at least 1
m/s (slow walk) will be generated. When the wind speed
is 1 m/s equation 7 reduces to:
Equation 8: WBGTid = 0.67Tpwb + 0.33Ta
Comparing equation 8 with equation 7 when the wind
speed is greater than 3 m/s, it can be seen that even for
higher wind speeds the WBGTid reduces by only about
than 6% of the 1 m/s value.
Tpwb is calculated from air temperature (Ta) and dew
point temperature (Td) by iteration using a formula derived from McPherson33):
Equation 933): 1556ed − 1.484edTpwb − 1556ew + 1.484ew
Tpwb + 1010(Ta − Tpwb) = 0
where ed = 6.106 exp (17.27Td/(237.3+Td)) (in hPa)
and ew = 6.106 exp (17.27Tpwb/(237.3+Tpwb)) (in hPa)
Hunter
Hunter28) used the principles of heat exchange to estimate globe temperature (Tg) by iteration of this formula:
Equation 1029): (1-αgs)SR(fdir/(4cos(z))+(1+αs)fdif ) + εa(1αgl)σTa4 = εσTg4 + 13.28v0.58(Tg-Ta)
The albedos (α) were assigned the following values:
globe shortwave αgs = 0.05, globe long-wave αgl = 0.05,
surrounds αs = 0.2. z is the zenith angle of the sun, fdir
Industrial Health 2012, 50, 267–278
WBGT CALCULATIONS FROM METEOROLOGICAL DATA
and fdif are the fractions of direct and diffuse radiation. v
is the wind speed in m/s (hence the different coefficient
before the v0.58 term from that in the formula in Hunter’s
paper28)). σ is the Stefan-Boltzmann constant. ε a is the
thermal emissivity of the air, and ε is the thermal emissivity of the globe = 0.95.
The two terms on the left of the equation represent the
short wavelength and long wavelength radiation absorbed
by the globe. The two terms on the right represent the
radiation emitted by the globe and the energy lost by
convection. The 13.28v0.58 is empirically derived34) for the
heat loss of the black globe from wind blowing over it.
To calculate Tnwb, Hunter28) used an empirical formula
derived for hot dry conditions in the USA (South Carolina). His formula (converted from degrees Fahrenheit to
Celsius) is:
Equation 1128): Tnwb = Tpwb+0.0117SR-0.233v+1.072
where v is wind speed in m/s and SR is solar radiation on
the wick (in W/m2) .
As the derivation of Tnwb (the main term in WBGT) is
empirical, this fails to meet one of the criteria for establishing the best formula. However, we will compare the
results of Hunter’s method with other methods to calculate
WBGT and comment on the closeness of fit of his formula.
ABM
The Australian Bureau of Meteorology (ABM) has
published on its website29) a simple formula for WBGT
that requires as input only water vapour pressure (ρ) and
air temperature (Ta).
Equation 1229): WBGT (°C) = 0.567 Ta + 0.393 ρ + 3.94
The website gives a standard physical science formula to
calculate ρ from RH (relative humidity) and Ta.
Equation 1329): ρ (hPa) = RH/100 × 6.105exp(17.27Ta/
(237.7+Ta))
While the ABM formula is easy to use, it proved difficult
to establish the origin of the formula as the reference
given35) does not have the formula. We tracked down the
original reference to an empirical regression fit of WBGT
vs temperature and humidity in a paper by Gagge and Nishi in 197636). Using their formula and converting to hPa
their WBGT relation is
Equation 1436): WBGTid = 0.567Ta + 0.216ρ +3.38
This is different to the formula quoted on the ABM web-
271
site. Further, this formula is empirically derived for indoor
conditions and does not include the effects of solar radiation and wind speed.
Thus, the ABM formula does not match our criteria
for validity. We include this formula in our comparisons
because other authors have used this formula to estimate
outdoor WBGT.
Tonouchi
Tonouchi30) calculated WBGT using the psychrometric
wet bulb temperature (Tpwb) instead of Tnwb. For the
solar component he used this empirically derived formula
for Tg:
Equation 1530): Tg (°C) = Ta + 0.0175SR − 0.208v
The radiation component used by Tonouchi was the solar
radiation incident on a flat horizontal plane instead of that
on a sphere.
This rendered the Tonouchi method as unsuitable as it
did not match our criteria in establishing a suitable formula. We will include this method in our comparisons as
an example of recent formulas derived empirically at one
location.
Liljegren and Gaspar
Liljegren et al.31) and Gaspar and Quintela32) used heat
exchange principles to calculate Tnwb and Tg. For Tnwb
they used a method similar to Bernard and for Tg they
refined the Hunter formula to include effects of different
amounts of radiation from the sky and the ground.
Liljegren
The Liljegren calculation for Tg includes both the direct
and diffuse components of sunlight so their method is
applicable for both sunny and cloudy conditions. Their
formula is not simple, but they make available a computer
program31) to calculate WBGT outdoors. The Liljegren
method meets all our criteria for outdoor WBGT calculations from meteorological data.
Liljegren’s formula assumes that when there is no solar
radiation the Tnwb equals Tpwb. As discussed earlier,
the Tpwb is defined as the wet bulb temperature without
sunlight exposure and a wind speed greater than 3 m/s. At
lower wind speeds the air around the wet bulb saturates
so preventing further evaporation (and hence further
cooling) resulting in Tnwb readings higher than Tpwb. As
indoor wind speeds are expected to be lower than 3 m/s,
Liljegren’s formula underestimates the Tnwb indoors. We
therefore decided to use the Liljegren formula only for
272
Fig. 1. The effect of solar radiation on calculated WBGT outdoors for various models.
Wind speed=1 m/s, humidity=55% and air temperature=30°C.
outdoor WBGT calculations.
Gaspar
The Gaspar 32) method requires the construction of
extensive formulas and a computer program for iteration.
This method is not easy to use. Further, the Gaspar Tg formula is only for clear skies and their measurements do not
consider the effect of direct and diffuse radiation. As the
high outdoor WBGT in tropical climates is often in humid
overcast conditions the Gaspar Tg does not meet our criteria of covering all common meteorological conditions.
Our comparative measurements
For indoor WBGT measurements we used a certified
WBGT meter (model QUESTemp 34 from Quest Technologies inc.). To measure the temperature and humidity
required for an indoor WBGT calculation we used a number of different temperature probes (eg Fluke 80T-150)
and humidity meters (eg UMP WS2015H). For outdoor
measurements we used the same Quest meter for WBGT
measurements. For humidity, wind speed, temperature and
radiation values required for the outdoor WBGT calculations we used a certified Campbell CR10 weather station
(see acknowledgements).
Results and Discussion: Comparing WBGT
Calculation Methods and Measurements
The previous section discussed various methods for de-
B LEMKE et al.
Fig. 2. The effect of humidity on calculated WBGT outdoors for
various models.
Solar radiation=500 W/m 2, wind speed=1 m/s, air temperature=30°C.
riving WBGT from readily available meteorological variables. Two methods met our validity criteria: the method
of Bernard for indoor WBGT and the method of Liljegren
for outdoor WBGT. While the following data compares
models, it should be noted that the various models have
been compared with actual data by the researchers who
developed the models. The Liljegren formula with Tg =
Ta could have been used for the indoors, but the Bernard
formula was a better fit to our criteria as it is much easier
to use.
Outdoor WBGT comparisons
Figures 1 through 4 compare the calculated outdoor
WBGTs using different methods for a range of climate
variables. The method by Hunter does not agree well with
the other methods. The empirical formula by Hunter 28) for
Tnwb (Equation 11) is problematic because it consistently
gives much higher Tnwb and WBGT values than the other
methods. Indeed, for the solar radiation on the white wetted wick alone, Equation 11 adds 8°C to the WBGT for
conditions of full sunlight. This is without including the
solar radiation acting on the globe. While Hunter tested
the accuracy of his formula, this was only on one day (early
summer) when the conditions (cloud cover) might have
been such that the calculations gave a good match with the
WBGT measurements.
As shown in Figs. 1 to 4, the published ABM formula29)
also does not agree with the other results. In particular, the
ABM formula does not vary with solar radiation nor wind
Industrial Health 2012, 50, 267–278
WBGT CALCULATIONS FROM METEOROLOGICAL DATA
273
Fig. 3. The effect of wind speed on calculated WBGT outdoors
for various models.
Solar radiation=500 W/m 2, Humidity=55% and air temperature=30°C.
Fig. 4. The effect of temperature on calculated WBGT outdoors
for various models.
Humidity=55%, Solar radiation=500 W/m 2 and wind speed=1 m/s.
speed as Gagge and Nishi 36) derived their formula for
indoor conditions that did not include those components.
The WBGT calculated via the Tonouchi formula30) is
close to, but lower than that of the other methods (Figs.
1 to 4). This is primarily because the Tonouchi formula
uses Tpwb rather than Tnwb in the calculation of WBGT,
which also explains why the influence of wind speed is
much reduced in the Tonouchi calculation (Fig. 3).
Liljegren31) made extensive comparisons between the
calculated outdoor WBGT and measured WBGT and
found that the difference between the two varied by less
than 1°C for 95% of the time, except in some locations
where they attributed the difference to equipment problems.
Figures 1 to 4 show that the methods of Liljegren31) and
Gaspar32) are in good agreement. Gaspar reported a small
but consistent overestimation in their calculations, which
could be attributed to very low wind speeds that prevented
accurate estimations of this parameter. The largest discrepancy between the results of Gaspar and Liljegren is in the
solar data. This is not surprising as the method of Gaspar
was only formulated for clear sky conditions.
slowly than a pyrometer to solar radiation. This coupled
with changing cloud cover so that one piece of equipment
reaches equilibrium while the other does not reach equilibrium makes direct comparisons difficult. Figure 5 shows
the actual outdoor WBGT (Quest) and the calculated outdoor WBGT (Liljegren) over a period of 4 days in Nelson,
New Zealand.
A difference of greater than 2°C can be seen at times.
However, the largest difference between the measured
and calculated values for all three days was between 11
a.m. and noon when the WBGT was rapidly increasing,
and presumably equilibrium conditions were not reached.
If the data between 11 a.m. and noon is removed then the
results comparing the measured and calculated outdoor
WBGT have a correlation coefficient of 98%.
The RMSE (root mean square error) between the
measured and calculated WBGT is 0.95°C. However the
RMSE between the calculated WBGT using the weather
station data and the calculated WBGT using the temperature and humidity data from the Quest equipment was
slightly higher at 0.97°C. Thus, differences in the temperature and humidity values between two nearby recorders resulted in about the same “error” as between the calculated
and measured outdoor WBGT values.
Field measurement comparisons outdoors
Accurate comparisons of calculated WBGT with measured values is very difficult in real outdoor situations.
Different equipment takes different times to reach equilibrium: for example, the WBGT black globe responds more
Indoor WBGT comparisons
Indoor WBGT readings are far more reproducible than
outdoor WBGT because the two major sources of varia-
274
B LEMKE et al.
Fig. 5. Measured outdoor WBGT compared to calculated WBGT
over 4 d in April 2010.
The data for the calculated values were taken from a nearby weather station. Location: near Nelson − north of the South Island of NZ.
Fig. 6. Indoor WBGT calculations by Liljegren,
ABM and Bernard.
Air temperature=30°C. Results for Gagge also included as a comparison with ABM.
Fig. 7. Measured indoor WBGT compared to calculated indoor WBGT.
Data from climate chamber. Temperature varied from 20 to 40°C, humidity varied from 30% to
90%. Wind speed approximately 0.5 m/s.
tion, cloud cover and wind speed, are eliminated. In our
calculations for the indoors, we assume there is some
movement by the workers that generates an apparent wind
speed of 1 m/s (slow walk).
Bernard27) extensively compared calculated WBGT with
indoor measurements and there is little point repeating
these. While the other researchers, whose methods have
been discussed in this paper have not carried out indoor
measurements, it is useful to compare Bernard’s results
with that of Liljegren and ABM as the latter was originally
derived for indoor conditions. Figure 6 shows this comparison. The ABM formula clearly gives an indoor WBGT
that is much too high. For comparison the figure also
shows results for a Gagge WBGT method (Equation 14)
on which the ABM formula is based.
While the method of Bernard and Liljegren are close
for indoor WBGT, it can be seen that the Liljegren results
are about 5% less than that using the Bernard method.
Liljegren did not test their formula indoors while Bernard
did extensive testing. This is one reason that we use the
Bernard formula for indoor WBGT calculations.
Comparison with our measurements
Figure 7 shows a comparison between typical indoor
WBGT measurements and WBGT calculations using
Equation 8. The WBGT readings were using a Quest
Industrial Health 2012, 50, 267–278
WBGT CALCULATIONS FROM METEOROLOGICAL DATA
Fig. 8. Indoor WBGT calculations by Bernard compared to measured indoor WBGT values.
Same experimental conditions as for Fig. 7.
WBGT meter which also recorded the relative humidity
and the air temperature, so it could be used to calculate
WBGT from the Liljegren formula. Figure 8 shows that
the agreement between theoretical and measured values is
within 0.5°C, except for a few points where it was identified that there was a rapid rise in humidity and the Tnwb
had not reached equilibrium. The difference between the
measured and calculated indoor WBGT has a RMSE =
0.47°C for all points, and RMSE = 0.40°C if the nonequilibrium points are removed.
Overall Discussion and Conclusions
A number of methods and calculation “models” have
been proposed for calculating the WBGT from standard
meteorological data. Some of these models are empirical (Tonouchi et al., ABM, Hunter, Bernard) and others
are based on thermodynamic heat exchange principles
(Dernedde and Gilbert, Liljegren et al., Gaspar and
Quintela). It is clear from our analysis that some WBGT
formulas derived from standard meteorological data are
better than others. We found that the Australian Bureau of
Meteorology formula quoted on their web site is incorrect.
Research based on the ABM formula (e.g. Hancock et
al.37)) needs to be re-evaluated.
While there are differences between the calculated and
measured WBGT using the “best models”, much of this
can be attributed to differences in the measurement of
temperature and humidity rather than errors in the formula.
It should be noted that WBGT measurements with special
equipment also have measurement errors25). These include
275
the fact that the standard 150 mm diameter black globe
takes up to 20 min to reach equilibrium so scattered cloud
will also cause errors. Non-standard 50 mm diameter
globes reach equilibrium more quickly, but give a slightly
smaller equilibrium temperature. Electronic wet bulb
temperature equipment may be less expensive but Tpwb is
measured rather than Tnwb. Good quality WBGT instruments are generally large and bulky which means they are
kept stationary rather than moving with the individual.
Hardcastle and Butler25) have compared a number of
commercial WBGT meters and found differences of between 1–2°C in their WBGT results. Hence the small difference between measured values and calculated WBGT
using the Liljegren method is less than the reported variability of measurement results.
Weather station and solar radiation data
Typical weather stations record hourly (or more or less
frequently) temperature, wind speed, and dew point. In
addition, recordings are usually made of atmospheric pressure, rainfall, wind direction and visibility. Historical data
going back decades or longer can be found from airports
at most large cities of the world. We used a database
including thousands of weather stations that is updated on
a daily basis by the US National Climatic Data Centre at
NOAA (National Oceanic and Atmospheric Administration, USA). Daily data (average, minimum, and maximum
24-h temperature; average wind speed; average humidity,
usually in the form of dew point Td; and some other variables) are available free from the NOAA web site 38) in the
GSOD (Global Summary of the Day) files. Hourly data is
available from the same web site or on CDs from NOAA
at a modest cost (about US$22 for 1 yr’s global data).
Unfortunately the NOAA data does not usually include
solar radiation, which is required for the most accurate
estimates of WBGT outdoors.
The best source of actual daily solar radiation is the satellite data from the NASA GEWEX site. This supplies the
average daily direct solar radiation for a 100 km × 100 km
area for any latitude and longitude. The current database
has data from 1983 to the present time. Useful data extracted from this site is the average daily short wave radiation (Average SR {day}) incident on that area of the globe
and the maximum possible direct solar radiation (Max SR
{day}), if there were no clouds.
The daily solar radiation data from NASA can be
converted to hourly SR data using a formula39) available
from the NOAA site40) that permits the calculation of the
maximum direct and diffuse solar radiation for clear skies
276
for each hour of the day for any day in the year and for
any latitude and longitude. This is the maximum solar
radiation that can fall on an area per hour: Max SR {hour}.
The calculated hourly average solar radiation for a particular day is then given by:
Equation 19: Average SR{hour} = Max SR{hour} × Average SR{day} / Max SR{day}
This hourly solar radiation has both the diffuse and the
direct component of solar radiation attenuated by the same
amount. In practice, this is incorrect as the clouds do not
attenuate the diffuse radiation to anywhere the same extent
as they attenuate the direct radiation. A better approximation is difficult to find as the proportion of direct and
diffuse radiation is very dependent on the type of cloud.
Zhang et al. 41) have used an empirical formula (including
cloud cover) developed by Watanabe et al.42) for China
and obtained good agreement with measurements.
It should be noted that for scattered clouds and rapidly
varying levels of heat radiation, the WBGT meter cannot
produce an accurate estimate of WBGT anyway as the
globe takes up to 20 min to reach equilibrium.
Future WBGT based on predicted climate change
Future WBGT trends related to climate change will
rely primarily on forecasted trends of daily temperature
and humidity43). Wind speeds are likely to change in less
predictable ways. Solar radiation on clear days will be the
same as now, but the future changes in cloud cover are difficult to forecast. For future outdoor WBGT predictions,
certain assumptions about wind speed and cloud cover
have to be made. The obvious assumption is that wind
speed and cloud cover do not change. As WBGT indoors
is not affected by solar radiation or wind speed then indoor
WBGT trends can be estimated from future climate modeling trends in “indoors” locations where indoor temperature
and humidity are the same as outdoors. This is a common
situation in low income communities and countries where
air conditioning is rare and windows are open or nonexistent. This applies to residential buildings as well as to
workshops and factories where poor people work44). This
allows us to estimate a WBGT field change45) using indoor
data and then super-impose that field change on actual
WBGT measurements for any location.
The ongoing climate change makes temporal and spatial
variations of workplace heat exposure into key public
health and occupational health issues in tropical and subtropical parts of the world. Our methods will improve the
analysis of these variations at a local, regional and country
B LEMKE et al.
level. New initiatives for occupational health management
will be necessary to protect the health and productivity of
working people46).
Our methods do not allow for WBGT calculation at a
specific work-place because workplace conditions can
be significantly different from conditions at the nearby
meteorological station. While WBGT meters are still best
for specific workplace conditions we are currently using
the methods outlined in this paper to enable us to calculate
specific workplace WBGTs from data obtained from cheap
data loggers that record only humidity and temperature.
We conclude that the Liljegren formula for WBGT
outdoors gave the most valid results, because it is based
on basic physics heat exchange formula and has been
extensively tested in the outdoors. For WBGT indoors
we concluded that the best is the Bernard formula with
Tg = Ta (assuming indoors with no local heat radiation
source), because it has been well tested by Bernard and
Pourmoghani.
These methods of calculating WBGT allow us to produce WBGT time trends going back to 1980 and beyond,
and to incorporate climate change predictions for a location in order to estimate future WBGT values.
Acknowledgements
The authors declare that we have no commercial interests in different kinds of heat measurements and analysis
or other conflicts of interest.
The research was supported by financial resources from
the National Centre for Epidemiology and Population
Health, Australian National University, Canberra, Australia. Some of these funds were from the Healthy Urban
Systems project co-sponsored by CSIRO, Australia.
The authors thank Tony Hewitt of Envirolink Ltd (Mapua
NZ) for access to local weather station data in Nelson.
References
1) d’Ambrosio Alfano FR, Palella BI, Riccio G (2011)
Thermal environment assessment reliability using
temperature − humidity indices. Ind Health 49, 95–106.
2) ISO (1989) Hot environments − Estimation of the heat
stress on working man, based on the WBGT-index (wet
bulb globe temperature). ISO Standard 7243. International
Standards Organization, Geneva.
3) UTCI Universal Thermal Climate Index http://www.utci.
org/. Accessed August 7, 2011.
4) ISO (1989) Hot environments − analytical determination
and interpretation of thermal stress using calculation of
Industrial Health 2012, 50, 267–278
WBGT CALCULATIONS FROM METEOROLOGICAL DATA
5)
6)
7)
8)
9)
10)
11)
12)
13)
14)
15)
16)
17)
18)
19)
20)
21)
22)
required sweat rate. ISO Standard 7933. International
Standards Organisation, Geneva.
Malchaire JBM (2006) Occupational heat stress assessment
by predicted heat strain model. Ind Health 44, 380–7.
Yokota M, Berglund L (2006) Initial capability decision
aid thermal prediction model and its validation. USARIEM
Technical Report T06-03.
Fiala D, Lomas K, Stohrer M (2001) Computer predictions
of human thermoregulatory and temperature responses to a
wide range of environmental conditions. Int J Biomet 45,
143–59.
Parsons K (2003) Human thermal environments. In: The
effects of hot, moderate and cold temperatures on human
health, comfort and performance. 2nd ed., Taylor &
Francis, London.
ILO (2001) Ambient factors in the workplace. International
Labour Organization (ILO) codes of practice. International
Labour Office, Geneva.
Parsons K (2006) Heat stress standard ISO 7243 and its
global application. Ind Health 44, 368–79.
Wakabayashi H, Wijayanto T, Lee JY, Hashiguchi N, Saat
M, Tochihara Y (2011) Comparison of heat dissipation
response between Malaysian and Japanese males during
exercise in humid heat stress. Int J Biometeorol 55, 509–17.
Yaglou CP, Minard D (1956) Prevention of heat casualties
at marine corps training centres. Armed Services Technical
Information Agency Document Service Center AD099920.
Stonehill RB, Keil PG (1961) Successful preventive
medical measures against heat illness at Lackland Air Force
Base. Am J Public Health Nations Health 51, 586–90.
Budd GM (2008) Wet-bulb globe temperature (WBGT)-its
history and its limitations. J Sci Med Sport 11, 20–32.
Brotherhood JR (2008) Heat stress and strain in exercise
and sport. J Sci Med Sport 11, 6–19.
Brake R, Bates GP (2002) A valid method for comparing
rational and empirical heat stress indices. Ann Occup Hyg
46, 165–74.
Park EK, Nag A, Ashtekar SP (2009) Thermal limits of men
in moderate to heavy work in tropical farming. Ind Health
47, 200–1.
Brake DJ, Bates GP (2002) Limiting work rate (Thermal
Work Limit) as an index of thermal stress. Appl Occup
Environ Hyg 17, 176–86 (ACGIH).
Parsons K (2011) Personal communication at Lund expert
panel meeting on International Guidelines on Indicators of
Occupational Heat Hazards Lund, Sweden July 2011.
Finucane E (2006) Definitions, conversions and calculations
for Occupational Safety and Health Professionals, 3rd Ed.,
Taylor and Francis, London.
Nakahama H (1999) Humidity measurement and
psychrometers in environmental testing equipment.
Environmental Testing Information: Technology Report
No.7 http://www.espec.co.jp/english/tech-info/tech_info/.
Accessed January 12, 2011.
Brice T, Hall T (2009) Wet-bulb calculator http://www.srh.
277
noaa.gov/epz/?n=wxcalc. Accessed July 16, 2011.
23) Bernard TE (2005) Psychrometrics and useful relationships
among heat stress measures. v1.0 http://www.health.usf.
edu/~tbernard. Accessed March 30, 2011.
24) Juang YJ, Lin YC (2007) The effect of thermal factors on
the measurement of wet globe temperature. J Occup Saf
Health 15, 191–203.
25) Hardcastle S, Butler K (2008) A comparison of globe,
wet and dry temperature and humidity measuring devices
available for heat stress assessment. 12th US/North
American Mine Ventilation Symposium, Wallace (Ed.).
26) Dernedde E, Gilbert D (1991) Prediction of wet-bulb globe
temperatures in aluminium smelters. Am Ind Hyg Assoc J
52, 120–6.
27) Bernard TE, Pourmoghani M (1999) Prediction of
workplace wet bulb global temperature. Appl Occup
Environ Hyg 14, 126–34.
28) Hunter C, Minyard O (1999) Estimating wet bulb globe
temperature using standard meteorological measurements.
Report WSRC-MS-99–00757. Westinghouse Savannah
River Company, Aiken.
29) ABM (2009) About the WBGT and apparent temperature
indices. Bureau of Meteorology, Commonwealth of
Australia, Melbourne, Website. http://www.bom.gov.au/
info/thermal_stress/. Accessed June 3, 2011.
30) Tonouchi M, Murayama K, Ono M (2006) WBGT forecast
for prevention of heat stroke in Japan. Sixth Symposium
on the Urban Environment. AMS Forum: Managing our
Physical and Natural Resources: Successes and Challenges
JP1.1.
31) Liljegren JC, Carhart R, Lawday P, Tschopp S, Sharp R
(2008) Modeling wet bulb globe temperature using standard
meteorological measurements. J Occup Environ Hyg 5,
645–55.
32) Gaspar AR, Quintela D (2009) Physical modelling of globe
and natural wet bulb temperatures to predict WBGT heat
stress index in outdoor environments. Int J Biometeorol 53,
221–30.
33) McPherson MJ (2008) Subsurface Ventilation and
Environmental Engineering, 2nd Ed., Chapter 17.
Physiological reactions to climatic conditions.
Mine Ventilation Services Inc., Clovis. http://www.
mvsengineering.com/index.php?cPath=25.
34) Kuehn LA, Stubbs RA, Weaver RS (1970) Theory of the
globe thermometer. J Appl Physiol 29, 750–7.
35) ACSM (1984) Prevention of thermal injuries during
distance running. American College of Sports Medicine,
Position Stand. Med J Aust 141, 876–9.
36) Gagge AP, Nishi Y (1976) Physical Indices of the thermal
environment. ASHRAE J 18, 47–51.
37) Hancock PA, Ross JM, Szalma JL (2007) A meta-analysis
of performance response under thermal stressors. Hum
Factors 49, 851–77.
38) NOAA (2011) National Oceanic and Atmospheric
Administration website. Global Summary of the day. http://
278
39)
40)
41)
42)
B LEMKE et al.
www.ncdc.noaa.gov/cgi-bin/res40.pl?page=gsod.html.
Accessed August 23, 2011.
Bird RE, Hulstrom RL (1981) A simplified clear-sky
model for the direct and diffuse insolation on horizontal
surfaces. US-SERI Technical Report TR-642–761, Golden,
Colorado.
Pelletier Greg (2002) solrad.xls (version 1.2) A solar
position and radiation calculator for Microsoft Excel/VBA
based on Bird and Hulstrom model. www.srrb.noaa.gov/
highlights/sunrise/azel.html. Accessed January 21, 2011.
Zhang Q, Huang J, Siwei L (2001) Development of Chinese
weather data for building energy calculations. Proc. 4th
Inter Conf on Indoor Air Quality, Ventilation and Energy
Conservation in Buildings. 1211, Changsha, Hunan.
Watanabe T, Urano Y, Hayashi H (1983) Procedures for
separating direct and diffuse insolation on a horizontal
surface and prediction of insolation on tilted surfaces. Trans
Arch Instit Jpn 330, 96–108.
43) IPCC (2007) Climate Change 2007: The physical
science basis. Summary for policymakers. Contribution
of working group I to the fourth assessment report. The
Intergovernmental Panel on Climate Change. http://www.
ipcc.ch/SPM2feb07.pdf.
44) Kjellstrom T (2009) Climate change exposures, chronic
diseases and mental health in urban populations–a threat to
health security, particularly for the poor and disadvantaged.
Technical report to the WHO Kobe Centre. Kobe, Japan.
World Health Organization, Geneva.
45) Tebaldi C, Knutti R (2007) The use of the multi-model
ensemble in probabilistic climate projections. Philos
Transact A Math Phys Eng Sci 365, 2053–75.
46) Kjellstrom T, Holmer I, Lemke B (2009) Workplace heat
stress, health and productivity − an increasing challenge for
low and middle-income countries during climate change.
Global Health Action. Special Volume, 46–51.
Industrial Health 2012, 50, 267–278